> ## Documentation Index
> Fetch the complete documentation index at: https://docs.axottle.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Axosun - review

> Built-in AI Operational Assistant in the Axottle Panel: Diagnostics, Reports, Incidents and Confirmation Actions.

> [!WARNING]
>
> Axosun is currently unavailable, and its return to service is not planned in the near future.

Axosun is a built-in operational assistant in the Axottle dashboard, which diagnoses infrastructure (nodes, Flow, health, Remnawave status), prepares structured reports with incident cards, and, once confirmed by the operator, performs actions from a fixed directory.

Axosun doesn't get constant interactive access to the shell. There's a special action in the directory called **to run a command on the** server to diagnose a specific node when regular checks aren't enough. For this, the panel shows the operator the exact command and reason for launching, and without confirmation, the command is not executed. Any action that isn't in the directory, the panel discards.

> [!NOTE]
>
> For whom: Administrator / SRE. Required rights: panel admin session. To perform actions - the appropriate permissions in the [policy Axosun](https://docs.axottle.com/en/axosun/permissions.md).

## How does Axosun help?

* **Diagnostics**: health checks, synthetic samples, DNS and TLS check, log analysis, comparison of expected and actual configuration. See also [Health System Review](https://docs.axottle.com/en/health/overview.md).
* \*\*Reports.\*\*A structured report on the status of the node, Flow, or the entire infrastructure with incident cards.
* **Incidents**Problems found are saved like cards with their own life cycle – they can be tracked, taken into work and closed.
* **Actions**: from restarting the service to mass operations and point diagnostics on the server, all taken from a fixed directory and validated.
* **Assistant Notes**: During launch, Axosun briefly explains what it does and why.

## Two modes of operation

\| Mode | How | Works
\|---|---|
\| **Classic** (default) | Axosun collects context, makes one model call and returns a structured report. The operator confirms the proposed actions after the report. |
\| **Agent** (optional, experimental) | Axosun works step-by-step: read-only checks are performed immediately during the dialogue, and before the destructive action, the launch is paused and a resolution card appears in the chat. |

Agent mode is enabled in the Axosun settings. It is applied to diagnostic launches whose model profile supports tool calling; in other cases, the launch is transparently performed in the classical mode with a note in the audit. Each agent launch is limited in the number of steps and time (no more than 8 steps and 6 minutes) - for more information in the section [Sessions and launches](https://docs.axottle.com/en/axosun/sessions-and-runs.md).

## How Axosun understands the request

The assistant classifies the operator’s message: a greeting, a question about the state of the infrastructure, diagnostics, requesting a report, handling an incident, repairing, direct action, scenario or inventory – and chooses the appropriate route of execution.

An explicit command like “restart WARP to `node-4`” follows a deterministic path without prior reporting: permission request → execution → mandatory verification of the result.

## Preliminary requirements

1. Connected LLM provider: at least one model profile with an API key is included - see [Connecting LLM provider](https://docs.axottle.com/en/axosun/configuration.md).
2. Configured Behavior Policy (by default, each action requires confirmation) – see [Law Model](https://docs.axottle.com/en/axosun/permissions.md).

## Section map

> [card] [Opportunities](https://docs.axottle.com/en/axosun/capabilities.md)
>
> Catalog of actions by group: diagnostics, nodes, Flow, Remnawave, incidents.

> [card] [The model of rights](https://docs.axottle.com/en/axosun/permissions.md)
>
> Modes of conduct, mandatory confirmations, access to Remnawave, approval leases.

> [card] [Data and privacy](https://docs.axottle.com/en/axosun/privacy.md)
>
> What is always cleaned, what is hidden by default and what is controlled by the administrator.

> [card] [Connecting an LLM provider](https://docs.axottle.com/en/axosun/configuration.md)
>
> Model profiles, API key storage, connection verification.

> [card] [Sessions and launches](https://docs.axottle.com/en/axosun/sessions-and-runs.md)
>
> Streaming progress, cancellation, confirmation of actions, limits.

> [card] [Safety locks](https://docs.axottle.com/en/axosun/safety.md)
>
> Protection against looping, preliminary check of requests, full audit.

> [card] [FAQ](https://docs.axottle.com/en/axosun/faq.md)
>
> Frequent questions about Axosun.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.